Question

The following query shows login information based on the userId. If a user closes their browser without logging off or the session expires, the logoff_date will remain null as in the example below.

userId    logon_date              logoff_date
1         2012-01-01 10:00:00     2012-01-01 12:00:00
1         2012-01-01 09:00:00     NULL

Because there is a newer logon_date of 2012-01-01 10:00:00, I know that the user must have killed the session for the login_date of 2012-01-01 09:00:00.

Here is my query:

SELECT userId, logon_date, logoff_date
FROM user_logon
WHERE user_id = 2

What I would like is to count only active sessions. In order to do this, I need to skip the rows where the logoff_date is missing if there is a newer row with the same userId.

Answer 1

How about this:

    SELECT l.*
      FROM user_logon l
INNER JOIN (SELECT MAX(logon_date) mdate,
                   user_id
              FROM user_logon
          GROUP BY user_id) x ON x.user_id = l.user_id
                             AND l.logon_date = x.mdate
     WHERE l.logoff_date IS NULL

?

PS: This query implies that for each date there is only one record for particular user